Enhancing POLB Mutation Classification with a Random Forest and PSO Hybrid Model

Abstract

Cancer formation and development are mainly attributable to DNA damage. When DNA is damaged, the rate of genetic mutations increases, hence the need for DNA repair mechanisms. Another essential element that should be further considered while discussing the base excision repair and its impact on the maintenance of genome stability is DNA polymerase β (Polβ), encoded by the POLB gene. This enzyme is used in humans to fix damaged DNA strings. The purpose of this study is to develop an accurate risk predictive model for cancer associated with a specific mutation. Hybrid Machine Learning (HML) classification algorithm has been applied to the POLB SNPs dataset. Random Forest combined with Particle Swarm Optimization (PSO) algorithm's hyperparameters to find and extract the best parameters. Through the models, the RF-PSO demonstrated superior performance, achieving an accuracy of 84.06%, precision of 84.49 %, sensitivity of 84.06%, specificity of 90.55%, and an F1 score of 83.81%.To verify the performance of the proposed algorithm, the accuracy of the suggested RF-PSO classifier model was compared with another state-of-the-art model classifier, Naive Bayes, K Nearest Neighbors, Stochastic Gradient Descent, Linear Discriminant Analysis, Gradient Boosting Machines, AdaBoost, Passive Aggressive, Extra Trees, and Hist Gradient Boosting. The results also proved the superior ability of the implemented RF-PSO model classifier in the classification to investigate the relationship between POLB gene variations and their potential role in cancer onset. providing a robust foundation for further clinical applications and which will further help in better cancer diagnosis and treatment of the disease.